X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=isamb%2Fisamb.c;h=1f57c7313550ed04cdf1dc32442a0ae614c461eb;hp=ab657d74b82c80b4f7b1630cce8e90ee6bcc1ea0;hb=49d0ee122a9f86ec2967b577dcc297c501785edd;hpb=b09c1d0aa414cb43fb667be54a29b3cc7279ea85

diff --git a/isamb/isamb.c b/isamb/isamb.c
index ab657d7..1f57c73 100644
--- a/isamb/isamb.c
+++ b/isamb/isamb.c
@@ -1,4 +1,4 @@
-/* $Id: isamb.c,v 1.72 2005-03-05 09:19:15 adam Exp $
+/* $Id: isamb.c,v 1.77 2005-04-15 10:47:49 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -72,7 +72,7 @@ struct ISAMB_head {
 #define ISAMB_PTR_CODEC 1
 
 struct ISAMB_cache_entry {
-    ISAMB_P pos;
+    ISAM_P pos;
     unsigned char *buf;
     int dirty;
     int hits;
@@ -102,7 +102,7 @@ struct ISAMB_s {
 };
 
 struct ISAMB_block {
-    ISAMB_P pos;
+    ISAM_P pos;
     int cat;
     int size;
     int leaf;
@@ -119,7 +119,7 @@ struct ISAMB_block {
 
 struct ISAMB_PP_s {
     ISAMB isamb;
-    ISAMB_P pos;
+    ISAM_P pos;
     int level;
     int maxlevel; /* total depth */
     zint total_size;
@@ -141,10 +141,10 @@ static void encode_ptr(char **dst, zint pos)
 
     while (pos > 127)
     {
-         *bp++ = 128 | (pos & 127);
+         *bp++ = (unsigned char) (128 | (pos & 127));
          pos = pos >> 7;
     }
-    *bp++ = pos;
+    *bp++ = (unsigned char) pos;
     *dst = (char *) bp;
 }
 #else
@@ -181,7 +181,7 @@ static void decode_ptr(const char **src, zint *pos)
 #endif
 
 ISAMB isamb_open(BFiles bfs, const char *name, int writeflag, ISAMC_M *method,
-                  int cache)
+		 int cache)
 {
     ISAMB isamb = xmalloc(sizeof(*isamb));
     int i, b_size = ISAMB_MIN_SIZE;
@@ -196,7 +196,7 @@ ISAMB isamb_open(BFiles bfs, const char *name, int writeflag, ISAMC_M *method,
     isamb->skipped_numbers = 0;
     isamb->returned_numbers = 0;
     for (i = 0; i<ISAMB_MAX_LEVEL; i++)
-      isamb->skipped_nodes[i] = isamb->accessed_nodes[i] = 0;
+	isamb->skipped_nodes[i] = isamb->accessed_nodes[i] = 0;
 
     assert(cache == 0);
     isamb->file = xmalloc(sizeof(*isamb->file) * isamb->no_cat);
@@ -266,9 +266,9 @@ ISAMB isamb_open(BFiles bfs, const char *name, int writeflag, ISAMC_M *method,
 	    decode_ptr(&src, &isamb->file[i].head.first_block);
 	    decode_ptr(&src, &isamb->file[i].head.last_block);
 	    decode_ptr(&src, &zint_tmp);
-	    isamb->file[i].head.block_size = zint_tmp;
+	    isamb->file[i].head.block_size = (int) zint_tmp;
 	    decode_ptr(&src, &zint_tmp);
-	    isamb->file[i].head.block_max = zint_tmp;
+	    isamb->file[i].head.block_max = (int) zint_tmp;
 	    decode_ptr(&src, &isamb->file[i].head.free_list);
 	}
         assert (isamb->file[i].head.block_size >= isamb->file[i].head.block_offset);
@@ -299,7 +299,7 @@ static void flush_blocks (ISAMB b, int cat)
     }
 }
 
-static int cache_block (ISAMB b, ISAMC_P pos, char *userbuf, int wr)
+static int cache_block (ISAMB b, ISAM_P pos, char *userbuf, int wr)
 {
     int cat = (int) (pos&CAT_MASK);
     int off = (int) (((pos/CAT_MAX) & 
@@ -432,7 +432,7 @@ void isamb_close (ISAMB isamb)
    * Reserve 5 bytes for large block sizes. 1 for small ones .. Number
    of items. We can thus have at most 2^40 nodes. 
 */
-static struct ISAMB_block *open_block(ISAMB b, ISAMC_P pos)
+static struct ISAMB_block *open_block(ISAMB b, ISAM_P pos)
 {
     int cat = (int) (pos&CAT_MASK);
     const char *src;
@@ -544,7 +544,7 @@ static void check_block (ISAMB b, struct ISAMB_block *p)
         char *startp = p->bytes;
         const char *src = startp;
         char *endp = p->bytes + p->size;
-        ISAMB_P pos;
+        ISAM_P pos;
 	void *c1 = (*b->method->codec.start)();
             
         decode_ptr(&src, &pos);
@@ -628,7 +628,7 @@ int insert_int (ISAMB b, struct ISAMB_block *p, void *lookahead_item,
     char *startp = p->bytes;
     const char *src = startp;
     char *endp = p->bytes + p->size;
-    ISAMB_P pos;
+    ISAM_P pos;
     struct ISAMB_block *sub_p1 = 0, *sub_p2 = 0;
     char sub_item[DST_ITEM_MAX];
     int sub_size;
@@ -839,6 +839,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
     int cut_item_size = 0;
     int no_items = 0;    /* number of items (total) */
     int no_items_1 = 0;  /* number of items (first half) */
+    int inserted_dst_bytes = 0;
 
     if (p && p->size)
     {
@@ -852,6 +853,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
         {
             const char *dst_item = 0; /* resulting item to be inserted */
             char *lookahead_next;
+	    char *dst_0 = dst;
             int d = -1;
             
             if (lookahead_item)
@@ -877,7 +879,6 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             else
                 dst_item = file_item_buf;
 
-	       
             if (!*lookahead_mode && d == 0)
             {
 		/* it's a deletion and they match so there is nothing to be
@@ -914,7 +915,8 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             {
 		/* we must move the lookahead pointer */
 
-                if (dst > maxp)
+		inserted_dst_bytes += (dst - dst_0);
+                if (inserted_dst_bytes >=  quater)
 		    /* no more room. Mark lookahead as "gone".. */
                     lookahead_item = 0;
                 else
@@ -966,17 +968,19 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
             }
         }
     }
-    maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
+
     /* this loop runs when we are "appending" to a leaf page. That is
        either it's empty (new) or all file items have been read in
        previous loop */
+
+    maxp = dst_buf + b->file[b->no_cat-1].head.block_max + quater;
     while (lookahead_item)
     {
         char *dst_item;
 	const char *src = lookahead_item;
         char *dst_0 = dst;
         
-	/* compare lookahead with max item */
+	/* if we have a lookahead item, we stop if we exceed the value of it */
         if (max_item &&
             (*b->method->compare_item)(max_item, lookahead_item) <= 0)
         {
@@ -1054,6 +1058,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
 	
 	/* first half */
         p->size = half1 - dst_buf;
+	assert(p->size <=  b->file[p->cat].head.block_max);
         memcpy (p->bytes, dst_buf, half1 - dst_buf);
 	p->no_items = no_items_1;
 
@@ -1069,6 +1074,7 @@ int insert_leaf (ISAMB b, struct ISAMB_block **sp1, void *lookahead_item,
         memcpy (first_dst, half2, dst - half2);
 
         (*sp2)->size = (first_dst - (*sp2)->bytes) + (dst - half2);
+	assert((*sp2)->size <=  b->file[p->cat].head.block_max);
 	(*sp2)->no_items = no_items - no_items_1;
         (*sp2)->dirty = 1;
         p->dirty = 1;
@@ -1102,7 +1108,7 @@ int insert_sub (ISAMB b, struct ISAMB_block **p, void *new_item,
                            sub_size, max_item);
 }
 
-int isamb_unlink (ISAMB b, ISAMC_P pos)
+int isamb_unlink (ISAMB b, ISAM_P pos)
 {
     struct ISAMB_block *p1;
 
@@ -1143,7 +1149,7 @@ int isamb_unlink (ISAMB b, ISAMC_P pos)
     return 0;
 }
 
-ISAMB_P isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
+void isamb_merge(ISAMB b, ISAM_P *pos, ISAMC_I *stream)
 {
     char item_buf[DST_ITEM_MAX];
     char *item_ptr;
@@ -1160,7 +1166,8 @@ ISAMB_P isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
             more =
                 (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
         }
-        return 1;
+	*pos = 1;
+        return;
     }
     item_ptr = item_buf;
     more = (*stream->read_item)(stream->clientData, &item_ptr, &i_mode);
@@ -1170,8 +1177,8 @@ ISAMB_P isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
         char sub_item[DST_ITEM_MAX];
         int sub_size;
         
-        if (pos)
-            p = open_block(b, pos);
+        if (*pos)
+            p = open_block(b, *pos);
         more = insert_sub (b, &p, item_buf, &i_mode, stream, &sp,
 			   sub_item, &sub_size, 0);
         if (sp)
@@ -1197,7 +1204,7 @@ ISAMB_P isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
             
             p2->size = dst - p2->bytes;
 	    p2->no_items = p->no_items + sp->no_items;
-            pos = p2->pos;  /* return new super page */
+            *pos = p2->pos;  /* return new super page */
             close_block(b, sp);
             close_block(b, p2);
 #if INT_ENCODE
@@ -1206,7 +1213,7 @@ ISAMB_P isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
         }
         else
 	{
-            pos = p->pos;   /* return current one (again) */
+            *pos = p->pos;   /* return current one (again) */
 	}
 	if (p->no_items == 0)
 	    must_delete = 1;
@@ -1216,13 +1223,12 @@ ISAMB_P isamb_merge (ISAMB b, ISAMC_P pos, ISAMC_I *stream)
     }
     if (must_delete)
     {
-	isamb_unlink(b, pos);
-	return 0;
+	isamb_unlink(b, *pos);
+	*pos = 0;
     }
-    return pos;
 }
 
-ISAMB_PP isamb_pp_open_x(ISAMB isamb, ISAMB_P pos, int *level, int scope)
+ISAMB_PP isamb_pp_open_x(ISAMB isamb, ISAM_P pos, int *level, int scope)
 {
     ISAMB_PP pp = xmalloc(sizeof(*pp));
     int i;
@@ -1264,12 +1270,12 @@ ISAMB_PP isamb_pp_open_x(ISAMB isamb, ISAMB_P pos, int *level, int scope)
     return pp;
 }
 
-ISAMB_PP isamb_pp_open (ISAMB isamb, ISAMB_P pos, int scope)
+ISAMB_PP isamb_pp_open (ISAMB isamb, ISAM_P pos, int scope)
 {
     return isamb_pp_open_x(isamb, pos, 0, scope);
 }
 
-void isamb_pp_close_x(ISAMB_PP pp, int *size, int *blocks)
+void isamb_pp_close_x(ISAMB_PP pp, zint *size, zint *blocks)
 {
     int i;
     if (!pp)
@@ -1312,7 +1318,7 @@ void isamb_pp_close (ISAMB_PP pp)
 }
 
 /* simple recursive dumper .. */
-static void isamb_dump_r (ISAMB b, ISAMB_P pos, void (*pr)(const char *str),
+static void isamb_dump_r (ISAMB b, ISAM_P pos, void (*pr)(const char *str),
 			  int level)
 {
     char buf[1024];
@@ -1341,7 +1347,7 @@ static void isamb_dump_r (ISAMB b, ISAMB_P pos, void (*pr)(const char *str),
 	else
 	{
 	    const char *src = p->bytes + p->offset;
-	    ISAMB_P sub;
+	    ISAM_P sub;
 
 	    decode_ptr(&src, &sub);
 	    p->offset = src - (char*) p->bytes;
@@ -1374,7 +1380,7 @@ static void isamb_dump_r (ISAMB b, ISAMB_P pos, void (*pr)(const char *str),
     }
 }
 
-void isamb_dump(ISAMB b, ISAMB_P pos, void (*pr)(const char *str))
+void isamb_dump(ISAMB b, ISAM_P pos, void (*pr)(const char *str))
 {
     isamb_dump_r(b, pos, pr, 0);
 }
@@ -1508,7 +1514,7 @@ static int isamb_pp_forward_on_leaf(ISAMB_PP pp, void *buf, const void *untilbuf
     }
 } /* forward_on_leaf */
 
-static int isamb_pp_climb_level(ISAMB_PP pp, ISAMB_P *pos)
+static int isamb_pp_climb_level(ISAMB_PP pp, ISAM_P *pos)
 { /* climbs higher in the tree, until finds a level with data left */
   /* returns the node to (consider to) descend to in *pos) */
     struct ISAMB_block *p = pp->block[pp->level];
@@ -1648,7 +1654,7 @@ static zint isamb_pp_forward_unode(ISAMB_PP pp, zint pos, const void *untilbuf)
     
 } /* forward_unode */
 
-static void isamb_pp_descend_to_leaf(ISAMB_PP pp, ISAMB_P pos,
+static void isamb_pp_descend_to_leaf(ISAMB_PP pp, ISAM_P pos,
 				     const void *untilbuf)
 { /* climbs down the tree, from pos, to the leftmost leaf */
     struct ISAMB_block *p = pp->block[pp->level];
@@ -1688,7 +1694,7 @@ static void isamb_pp_descend_to_leaf(ISAMB_PP pp, ISAMB_P pos,
 
 static int isamb_pp_find_next_leaf(ISAMB_PP pp)
 { /* finds the next leaf by climbing up and down */
-    ISAMB_P pos;
+    ISAM_P pos;
     if (!isamb_pp_climb_level(pp, &pos))
         return 0;
     isamb_pp_descend_to_leaf(pp, pos, 0);
@@ -1697,7 +1703,7 @@ static int isamb_pp_find_next_leaf(ISAMB_PP pp)
 
 static int isamb_pp_climb_desc(ISAMB_PP pp,  const void *untilbuf)
 { /* climbs up and descends to a leaf where values >= *untilbuf are found */
-    ISAMB_P pos;
+    ISAM_P pos;
 #if ISAMB_DEBUG
     struct ISAMB_block *p = pp->block[pp->level];
     yaz_log(YLOG_DEBUG, "isamb_pp_climb_desc starting "
@@ -1798,7 +1804,7 @@ void isamb_pp_pos(ISAMB_PP pp, double *current, double *total)
     assert(current);
     assert(p->leaf);
 
-    *total = pp->block[0]->no_items;
+    *total = (double) (pp->block[0]->no_items);
     *current = (double) pp->returned_numbers;
 #if ISAMB_DEBUG
     yaz_log(YLOG_LOG, "isamb_pp_pos returning: cur= %0.1f tot=%0.1f rn="
@@ -1817,7 +1823,7 @@ int isamb_pp_forward2(ISAMB_PP pp, void *buf, const void *untilb)
 again:
     while (p->offset == p->size)
     {
-        ISAMB_P pos;
+        ISAM_P pos;
 #if INT_ENCODE
 	const char *src_0;
 	void *c1;